dsh-plugin-lookatstudy 0.25.0 → 0.25.2

package/lib/index.mjs

@@ -266,7 +266,6 @@ const ZH = {
 	"chat.reasoning": "思考过程（{n} 字）",
 	"chat.think": "思考",
 	"langview.original": "原文",
-	"langview.bilingual": "对照",
 	"langview.translation": "译文",
 	"msg.speak": "朗读",
 	"note.pin": "置顶笔记",
@@ -638,7 +637,6 @@ const EN = {
 	"chat.reasoning": "Thinking ({n} chars)",
 	"chat.think": "Think",
 	"langview.original": "Original",
-	"langview.bilingual": "Both",
 	"langview.translation": "Translation",
 	"msg.speak": "Speak",
 	"note.pin": "Pin note",
@@ -838,21 +836,6 @@ function escapeHtml(text) {
 function inline(escaped) {
 	return escaped.replace(/`([^`]+)`/g, "<code>$1</code>").replace(/\*\*([^*]+)\*\*/g, "<strong>$1</strong>").replace(/\*([^*]+)\*/g, "<em>$1</em>").replace(/!\[([^\]]*)\]\((https?:\/\/[^)\s]+|data:image\/[^)\s]+|\/lookatstudy\/api\/attachment\/[^)\s]+)\)/g, (_m, alt, src) => `<img src="${src}" alt="${alt}" loading="lazy">`).replace(/\[([^\]]+)\]\((https?:\/\/[^)\s]+)\)/g, "<a href=\"$2\" target=\"_blank\" rel=\"noreferrer\">$1</a>");
 }
-/** Interleave one body with its translation, paragraph by paragraph (upstream
-* translation-layout "microsoft" semantics: original paragraph then its
-* translation as a quote, tight pairing for bilingual reading; length
-* mismatches append the remainder in order). Pure. */
-function renderBilingual(bodyMd, translationMd) {
-	const par = (md) => md.split(/\n\s*\n/).map((p) => p.trim()).filter((p) => p !== "");
-	const a = par(bodyMd);
-	const b = par(translationMd);
-	const out = [];
-	for (let i = 0; i < Math.max(a.length, b.length); i++) {
-		if (a[i] !== void 0) out.push(a[i]);
-		if (b[i] !== void 0) out.push(`> ${b[i].split("\n").join("\n> ")}`);
-	}
-	return out.join("\n\n");
-}
 /** True when the line opens a GFM table row (pipes with a delimiter row next). */
 function isTableRow(line) {
 	return line.trim().startsWith("|") && line.trim().endsWith("|") && line.includes("|", 1);
@@ -3138,7 +3121,6 @@ function workbenchState(state, now) {
 			markdown: ref.lesson.body,
 			speechText: normalizeSpeechText(normalizeMathNotation(ref.lesson.body)),
 			...ref.lesson.translation !== void 0 ? {
-				bilingualHtml: renderMarkdown(normalizeMathNotation(renderBilingual(ref.lesson.body, ref.lesson.translation))),
 				translationHtml: renderMarkdown(normalizeMathNotation(ref.lesson.translation)),
 				translationLang: ref.lesson.translationLang ?? ""
 			} : {}
@@ -6599,6 +6581,75 @@ function validateDesign(design, validFiles) {
 		droppedLessons
 	};
 }
+/**
+* Heading similarity — the translation misalignment guard. Ported from
+* upstream import-pipeline.ts (their audit F16, 2026-09-13): ordinal
+* alignment assumes both heading lists share structure; a machine-translated
+* file with a dropped header shifts every ordinal, and A's translation would
+* silently land on B's node — overlap too low skips the segment (宁缺毋错).
+* Empty titles cannot be checked and pass conservatively.
+*
+* 0.25.2 (owner real-import catch): upstream's word overlap normalizes on
+* whitespace, and a CJK title is ONE token — paraphrased Chinese headings
+* (课程总览与环境搭建指南 vs 课程总览) could never overlap and the guard
+* skipped every non-verbatim segment (38/110 paired on the owner's import;
+* the 38 were exact matches). CJK strings now compare by character
+* containment first, then character-bigram overlap ≥ 0.3.
+*/
+function headingsSimilar(a, b) {
+	const norm = (s) => s.toLowerCase().replace(/[^\p{L}\p{N}]+/gu, " ").trim().split(/\s+/).filter(Boolean);
+	const A = norm(a);
+	const B = norm(b);
+	if (A.length === 0 || B.length === 0) return true;
+	if (A.join(" ") === B.join(" ")) return true;
+	const han = (s) => s.toLowerCase().replace(/[^\p{Script=Han}\p{N}]/gu, "");
+	const ca = han(a);
+	const cb = han(b);
+	if (ca.length >= 2 || cb.length >= 2) {
+		if (ca === "" || cb === "") return true;
+		if (ca.includes(cb) || cb.includes(ca)) return true;
+		const bigramsOf = (s) => {
+			const out = /* @__PURE__ */ new Set();
+			const chars = Array.from(s);
+			for (let i = 0; i + 1 < chars.length; i++) out.add(chars[i] + chars[i + 1]);
+			return out;
+		};
+		const ba = bigramsOf(ca);
+		const bb = bigramsOf(cb);
+		let hit = 0;
+		for (const g of ba) if (bb.has(g)) hit++;
+		return hit / Math.max(1, ba.size) >= .3;
+	}
+	const setB = new Set(B);
+	return A.filter((w) => setB.has(w)).length / A.length >= .3;
+}
+/**
+* Replace a translation body's image references BY POSITION with the
+* original lesson's images (upstream method, import-pipeline.ts): translated
+* figure i ← original figure i (the original's refs are already absolute
+* CDN URLs or inlined data URLs); extras beyond the original count drop
+* (machine translation adding figures is rare — dropping is cleanest);
+* HTML imgs convert to markdown. One unified regex pass so HTML→markdown
+* conversion is never re-scanned with a shared index.
+*/
+function replaceImagesByPosition(content, originalImgs) {
+	let imgIdx = 0;
+	return content.replace(/!\[([^\]]*)\]\(([^)]+)\)|<img[^>]+src=["']([^"']+)["'][^>]*>/gi, (match, mdAlt, _mdSrc, htmlSrc) => {
+		if (imgIdx >= originalImgs.length) return "";
+		const replacement = originalImgs[imgIdx++];
+		if (htmlSrc !== void 0) return `![${/alt=["']([^"']*)["']/i.exec(match)?.[1] ?? ""}](${replacement})`;
+		return `![${mdAlt}](${replacement})`;
+	});
+}
+/** The original lesson body's image refs in order (the position-mapping source). */
+function originalImagesOf(body) {
+	const out = [];
+	for (const m of body.matchAll(/!\[([^\]]*)\]\(([^)]+)\)|<img[^>]+src=["']([^"']+)["'][^>]*>/gi)) {
+		const src = m[2] ?? m[3] ?? "";
+		if (src !== "") out.push(src);
+	}
+	return out;
+}
 /** GitHub-style anchor slug for ParsedLesson/ParsedSection anchors. */
 function slugAnchor(title) {
 	const slug = title.toLowerCase().replace(/[^a-z0-9]+/g, "-").replace(/^-+|-+$/g, "");
@@ -6635,13 +6686,6 @@ function buildCourseFromDesign(courseTitle, validated, contents, translations) {
 			const titleIndex = lesson.anchor === null ? -1 : findTitleIndex(headings, lesson.anchor);
 			const isFirst = !firstLessonSeen.has(lesson.file);
 			firstLessonSeen.add(lesson.file);
-			const pair = translations?.get(lesson.file);
-			let translation;
-			if (pair !== void 0) {
-				const tHeadings = extractHeadings(pair.content);
-				const tIndex = lesson.anchor === null ? -1 : findTitleIndex(tHeadings, lesson.anchor);
-				translation = sliceLessonBody(pair.content, tHeadings, tIndex, isFirst);
-			}
 			const body = sliceLessonBody(content, headings, titleIndex, isFirst);
 			if (body.length > 8e3 && splittableHeadingCount(headings, titleIndex) > 0) pacingViolations.push({
 				title: lesson.title,
@@ -6649,6 +6693,17 @@ function buildCourseFromDesign(courseTitle, validated, contents, translations) {
 				chars: body.length,
 				splittable: splittableHeadingCount(headings, titleIndex)
 			});
+			const pair = translations?.get(lesson.file);
+			let translation;
+			if (pair !== void 0) {
+				const tHeadings = extractHeadings(pair.content);
+				let tIndex = titleIndex;
+				let keep = true;
+				if (lesson.anchor === null || titleIndex === -1) tIndex = -1;
+				else if (titleIndex >= tHeadings.length) keep = false;
+				else if (!(tHeadings.length === headings.length && tHeadings.every((h, i) => h.level === headings[i].level)) && !headingsSimilar(lesson.title, tHeadings[titleIndex].title)) keep = false;
+				if (keep) translation = replaceImagesByPosition(sliceLessonBody(pair.content, tHeadings, tIndex, isFirst), originalImagesOf(body));
+			}
 			return {
 				title: lesson.title,
 				anchor: slugAnchor(lesson.title),
@@ -7710,7 +7765,7 @@ ${p.text}`));
 					translations: {
 						type: "integer",
 						required: true,
-						description: "Lessons that landed with a paired translation (the interface-language mirror existed and its anchor matched) — their teach tab offers 原文/对照/译文."
+						description: "Lessons that landed with a paired translation (the interface-language mirror existed and its anchor matched) — their teach tab offers 原文/译文."
 					},
 					created: {
 						type: "boolean",
@@ -7725,7 +7780,7 @@ ${p.text}`));
 			},
 			render: (_args, value) => [{
 				type: "text",
-				text: value.created === false ? `“${value.title}” (id ${value.courseId}) was ALREADY imported — the design did not create a new course. Continue with the existing course, or delete it first (study_delete_course) to rebuild from this design.` : `Imported designed course “${value.title}” (${value.sections} sections, ${value.lessons} lessons${value.droppedLessons > 0 ? `, ${value.droppedLessons} hallucinated lesson(s) dropped` : ""})${typeof value.translations === "number" && value.translations > 0 ? ` — ${value.translations} lesson(s) carry the learner's-language translation (原文/对照/译文 in the teach tab)` : ""}${typeof value.languageTarget === "string" ? ` — language course (${value.languageTarget}): keep reading passages, example sentences, and quiz language material in the target language` : ""}. First lesson: “${value.firstLessonTitle}” (id ${value.firstLessonId}). Present the course map to the learner.`
+				text: value.created === false ? `“${value.title}” (id ${value.courseId}) was ALREADY imported — the design did not create a new course. Continue with the existing course, or delete it first (study_delete_course) to rebuild from this design.` : `Imported designed course “${value.title}” (${value.sections} sections, ${value.lessons} lessons${value.droppedLessons > 0 ? `, ${value.droppedLessons} hallucinated lesson(s) dropped` : ""})${typeof value.translations === "number" && value.translations > 0 ? ` — ${value.translations} lesson(s) carry the learner's-language translation (原文/译文 in the teach tab)` : ""}${typeof value.languageTarget === "string" ? ` — language course (${value.languageTarget}): keep reading passages, example sentences, and quiz language material in the target language` : ""}. First lesson: “${value.firstLessonTitle}” (id ${value.firstLessonId}). Present the course map to the learner.`
 			}]
 		},
 		async execute(args, exec) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dsh-plugin-lookatstudy",
-  "version": "0.25.0",
+  "version": "0.25.2",
   "packageManager": "pnpm@11.7.0",
   "description": "Turn any markdown, local folder, or GitHub learning repo into a guided course inside DeepSeek Harness: gated skill-tree progression, BKT mastery tracking, SM-2 spaced repetition.",
   "type": "module",